LAMOL: LAnguage MOdeling for Lifelong Language Learning

Fan-Keng Sun, Cheng-Hao Ho, Hung-yi Lee

arXiv (Cornell University) · 2019 · 84 citations · 24 references

DOIFull text

Open access

Concepts

TL;DR

Lifelong learning has largely focused on images or games, leaving language underexplored. The authors introduce LAMOL, a language‑model‑based approach for lifelong language learning. LAMOL learns to solve tasks and generate pseudo‑samples of previous tasks, replaying them during training of new tasks without extra memory or model capacity. LAMOL prevents catastrophic forgetting, performs five diverse language tasks sequentially with a single model, outperforms prior methods by a considerable margin, and is only 2–3 % worse than multitasking, the typical LLL upper bound. The source code is available at https://github.com/jojotenya/LAMOL.

Abstract

Most research on lifelong learning applies to images or games, but not language. We present LAMOL, a simple yet effective method for lifelong language learning (LLL) based on language modeling. LAMOL replays pseudo-samples of previous tasks while requiring no extra memory or model capacity. Specifically, LAMOL is a language model that simultaneously learns to solve the tasks and generate training samples. When the model is trained for a new task, it generates pseudo-samples of previous tasks for training alongside data for the new task. The results show that LAMOL prevents catastrophic forgetting without any sign of intransigence and can perform five very different language tasks sequentially with only one model. Overall, LAMOL outperforms previous methods by a considerable margin and is only 2-3% worse than multitasking, which is usually considered the LLL upper bound. The source code is available at https://github.com/jojotenya/LAMOL.

References

24